JP2020159301A - Engine controller - Google Patents

Abstract

To provide an engine controller that when the occurrence of failure due to air-fuel ratio deviation between cylinders continues, even if long-time learning by sub-feedback control accumulates, can correct an air-fuel ratio to a lean side to suppress exhaust gas.SOLUTION: A sub-feedback control means 32 adjusts a lean correction value, which is a value of the real-time correction amount at the time of lean correction, in a plurality of steps or steplessly, based on an integrated value of a lean residence time.SELECTED DRAWING: Figure 3

Description

The present invention relates to an engine control device.

Generally, an exhaust purification device equipped with a catalyst is provided to purify the exhaust gas generated by the engine. This catalyst is designed to function effectively when the engine is burned under conditions close to the stoichiometric air-fuel ratio (stoichiometry) in which oxygen and fuel react in just proportion. In order to control the air-fuel ratio, which is the ratio of oxygen and fuel actually burned in the engine, to approach the theoretical air-fuel ratio, the first air-fuel ratio sensor (linear air-fuel ratio) is upstream from the exhaust purification device in the exhaust passage of the engine. A sensor / Linear Air Fuel Radio Sensor) is provided, and main feedback control for controlling the fuel supply amount to the engine is performed so that the value of this air-fuel ratio sensor matches the target air-fuel ratio.

However, the value obtained from the first air-fuel ratio sensor located upstream of the exhaust gas purification device fluctuates drastically. In order to correct this, sub-feedback control that complements the control value performed by the main feedback control is also performed based on the value of the second air-fuel ratio sensor provided downstream of the exhaust gas purification device.

By the way, generally, an engine having a plurality of combustion chambers of two or more cylinders is used, and exhaust gas exhausted from those combustion chambers is collected and sent to an exhaust purification device. The first air-fuel ratio sensor is provided on the downstream side of the point where the exhaust paths from those combustion chambers are merged.

As an example of a control device that provides such feedback, in Patent Document 1, a catalytic converter is installed as an exhaust purification device in the exhaust passage of a 4-cylinder engine, and a catalytic converter sensor and a catalytic converter are located on the upstream and downstream sides of the catalytic converter. A control device provided with a post-catalyst sensor is described.

Japanese Unexamined Patent Publication No. 2013-19334

Air-fuel ratio deviation occurs between multiple cylinders When an air-fuel ratio deviation failure occurs between cylinders, the output of the first air-fuel ratio sensor is generally on the rich side, which is overfueled, even when operating at the theoretical air-fuel ratio. It tends to shift. When the output of the first air-fuel ratio sensor shifts to the rich side, the main feedback control tries to correct this and corrects the air-fuel ratio to the lean side where the fuel is insufficient, so the catalyst of the exhaust purification device operates in a lean atmosphere. Become. At this time, the output of the second air-fuel ratio sensor is also lean. When that state continues, sub-feedback control is performed so that the air-fuel ratio targeted by the main feedback control is corrected to the rich side according to the time spent in the lean, and the deterioration of the exhaust gas is suppressed. This is called long-time learning.

However, the way the exhaust gas hits the first air-fuel ratio sensor located closest to the downstream side changes depending on the cylinder and operating point that caused the air-fuel ratio deviation between cylinders, so long-time learning is always controlled in the same way. It is not possible to do it properly. In some cases, the long-time learning works too much, and even though the air-fuel ratio deviation failure between cylinders continues, a learning return may occur in which the failure determination is not established. In addition, since long-time learning is corrected to the rich side according to the time spent in lean, if the staying time is accumulated too much, the overcorrection becomes too rich as a whole. In the case of overcorrection, even if the state should be corrected to the lean side, the air-fuel ratio may not be corrected by the normal sub-feedback control alone because the bias is too rich.

Therefore, the present invention suppresses exhaust gas by making it possible to correct to the lean side even if long-time learning by sub-feedback control is accumulated when the occurrence of air-fuel ratio deviation failure between cylinders continues. With the goal.

In order to solve the above problems, the present invention is provided in an injection device for supplying fuel to an engine having a plurality of cylinders and an exhaust passage drawn from the plurality of cylinders, and the air-fuel ratio in the exhaust passage is adjusted. The first air-fuel ratio detecting means to be detected, the exhaust purification device provided on the downstream side of the first air-fuel ratio detecting means, and the exhaust passage provided on the downstream side of the exhaust purification device are provided in the exhaust passage. Based on the information obtained by the second air-fuel ratio detecting means for detecting the air-fuel ratio and the first air-fuel ratio detecting means, the air-fuel ratio of the fuel produced by the injection device is set to a preset target air-fuel ratio. If the information obtained by the main feedback control means that feedback-controls the injection amount and the second air-fuel ratio detecting means is leaner than the lean predetermined value set on the lean side, the target air-fuel ratio is richly corrected and the rich side. If it is on the rich side of the rich predetermined value set in, the real-time correction amount for lean-correcting the target air-fuel ratio and the lean retention obtained based on the information obtained by the second air-fuel ratio detecting means for the real-time correction amount. A sub-feedback control means for setting a long time correction amount for correction according to an integrated value of time is provided, and when an air-fuel ratio deviation between cylinders occurs, the air-fuel ratio shift occurs between the plurality of cylinders. The sub-feedback control means employs an engine control device that adjusts the lean correction value of the real-time correction amount at the time of lean correction in a plurality of steps or steplessly based on the integrated value of the lean residence time. did.

Here, the integrated value of the lean residence time can adopt a configuration that is calculated based on the total time that the residence determination value obtained by filtering the real-time correction amount stays at a predetermined value on the rich side. Since the real-time correction amount fluctuates so as to immediately react and correct the noise and extreme fluctuations detected by the second air-fuel ratio detecting means, the time when the real-time correction amount becomes the rich side is simply simplified. If it is accumulated to, the integrated value will be added even in a situation where it should not be determined that the correction is made to the rich side. The integration is performed only when it should be determined that the second air-fuel ratio detecting means clearly shows a value biased toward the lean side and corrects it toward the rich side by using the retention judgment value whose fluctuation is rounded by filtering. The values will be added.

Further, it is possible to adopt a configuration in which the lean correction value is held as a reserve value and the standby correction amount control means for applying the reserve value at the time of subsequent acceleration is provided. By setting the lean correction value, which is the value when changing the real-time correction amount to the lean side, as a preparation value in advance, the lean correction can be performed without newly incorporating the timing for changing the real-time correction amount in the processing flow. Even if there is a bias toward the rich side due to long-time learning when it becomes necessary, it will be possible to sufficiently correct it toward the lean side.

Further, the long time correction amount is not reflected when the integrated value of the lean residence time is less than the predetermined variation determination value, and the correction to the rich side is reflected when the variation determination value is exceeded. Can be adopted. The characteristics of each engine vary, and the integrated value of the lean residence time may be added even if it is not particularly deteriorated or failed due to the influence of this variation. The effects of such individual differences can be eliminated and feedback can only be affected in situations where there is a clear tendency to lead to deterioration or failure.

Furthermore, a configuration can be adopted in which a residence time subtracting means is executed in which the integrated value of the lean residence time is subtracted according to the residence time at which the residence determination value stays on the lean side from the lean determination value. Depending on the situation, the air-fuel ratio deviation between cylinders may be corrected, and in such a case, the situation biased toward the rich side can be corrected by long-time learning. On the other hand, when adopting a configuration in which the residence time subtraction means can be executed, a configuration in which the learning return suppression means is executed can be adopted in which the lean determination value is lowered according to the integrated value of the lean residence time. In order to perform deterioration judgment and failure judgment according to such residence time, it is possible that the second air-fuel ratio detecting means only temporarily shows a value that is biased toward the rich side and overturns this judgment. It becomes inconvenient when a serious failure occurs. For this reason, by lowering the lean judgment value, which is the threshold value for starting learning return, according to the integrated value of the lean residence time, it is easy to increase the lean residence time and when there is a high possibility that the failure is serious. Improve the accuracy of the judgment by preventing the learning return.

By controlling the engine with the control device according to the present invention, when the air-fuel ratio deviation between cylinders fails and the situation of correcting the rich tendency continues, even if the long time learning is accumulated and overcorrected, the exhaust is discharged. Gas can be suppressed.

An overall view schematically showing an engine control device showing an example of one embodiment of the present invention. Flowchart showing the outline of the control contents Transition diagram showing the relationship of specific control information

An embodiment of the present invention will be described with reference to the drawings. FIG. 1 is an overall view showing the configuration of the control device of the engine 1 of the present invention.

The engine 1 is a four-cylinder engine for automobiles. As shown in FIG. 1, four cylinders 2 are provided in parallel, and are drawn out from an intake passage 4 and an exhaust port (not shown) leading to an intake port (not shown) for feeding an air-fuel mixture into each cylinder 2. It is provided with an exhaust passage 5, an in-cylinder injection device 10, and the like.

Note that FIG. 1 shows only the members and means directly related to the present invention, and the other members and the like are not shown. Further, although the drawing shows an example in which four cylinders 2 are provided, the engine may have two cylinders, and the cylinders may not be arranged in a single row.

A throttle valve 3 for adjusting the flow path area is provided upstream of the intake passage 4 before branching, so that the intake amount can be adjusted.

The first air-fuel ratio detecting means (O ₂ sensor) 12 is attached to the place where the individual exhaust passages 5 meet. At the end of the merged exhaust passage 11, an exhaust purification unit 13 provided with a catalyst or the like for removing nitrogen oxides or the like in the exhaust toward the downstream side, and a second air-fuel ratio detecting means (rear O) further downstream thereof. ₂ sensor) 14 is attached, and a muffler 15 and the like are further provided on the downstream side thereof.

The equipment necessary for the operation of the engine, including the throttle valve 3 and the in-cylinder injection device 10, is controlled by the electronic control unit 30 included in the vehicle equipped with the engine 1, respectively. Further, various information from the first air-fuel ratio detecting means 12, the second air-fuel ratio detecting means 14, and the like is transmitted to the electronic control unit 30.

The electronic control unit 30 controls the throttle valve 3 to adjust the intake amount of the intake passage 4. Further, the amount of fuel injected from the in-cylinder injection device 10 to the cylinder 2 is controlled. By these adjustments, the air-fuel ratio can be adjusted toward the stoichiometric air-fuel ratio, which is the stoichiometric air-fuel ratio, the rich side with a large amount of fuel, or the lean side with a small amount of fuel.

The injected fuel is mixed with the intake air from the intake passage 4, and an air-fuel mixture is formed in the cylinder 2. By igniting a spark plug (not shown) provided in the cylinder 2, the air-fuel mixture is burned to generate engine torque. The exhaust gas of each cylinder 2 is discharged to the exhaust passage 5, merges with the exhaust passage 11, is purified by the exhaust purification unit 13, and is discharged through the muffler 15.

The electronic control unit 30 performs main feedback control so that the air-fuel ratio becomes the target air-fuel ratio based on the information of the first air-fuel ratio detecting means 12 according to the driving condition of the vehicle. This control is performed by the main feedback control means 31 included in the electronic control unit 30 adjusting the fuel injection amount by the in-cylinder injection device 10.

Since the first air-fuel ratio detecting means 12 receives and detects the gas exhausted from the parallel exhaust passage 5, the influence differs for each cylinder. Further, the air-fuel ratio has a habit for each cylinder 2, and not all cylinders 2 show the same behavior. For various reasons other than these, the main feedback control by the first air-fuel ratio detecting means 12 is not sufficient, and the sub-feedback control for adjusting the target air-fuel ratio targeted by the main feedback control is performed. This control is performed by the sub-feedback control means 32 included in the electronic control unit 30 by changing the set value inside the electronic control unit 30.

The target air-fuel ratio is set by combining the real-time correction amount and the long-time correction amount. The real-time correction amount is immediately reflected based on the value of the second air-fuel ratio detecting means 14. While the value of the second air-fuel ratio detecting means 14 is within a predetermined range, the real-time correction amount is set to ± 0. This real-time correction amount is richly corrected while it is on the lean side from the predetermined lean value set in advance on the lean side. Lean correction is performed while the rich side is higher than the rich predetermined value set in advance on the rich side. Both rich correction and lean correction are changed to predetermined values. The value for rich correction is called the rich correction value, and the value for lean correction is called the lean correction value. Of these, the rich correction value may basically be a default value. On the other hand, the electronic control unit 30 according to the present invention adjusts the lean correction value in a plurality of steps or steplessly. That is, the lean correction value is variable.

The integrated value of the lean residence time is used as a criterion for changing the lean correction value. The integrated value of the lean residence time is based on the total sum of counting the times when the value of the second air-fuel ratio detecting means 14 is on the lean side of the predetermined lean value. However, if an attempt is made to count all the times when the value of the second air-fuel ratio detecting means 14 is on the lean side, even the momentary fluctuation of the value of the second air-fuel ratio detecting means 14 is picked up, which is rather inaccurate. In some cases. Therefore, the period in which the real-time correction amount reflected based on the value of the second air-fuel ratio detecting means 14 is set to the rich correction value (that is, the value of the second air-fuel ratio detecting means 14 is lean side from the lean predetermined value). It is desirable to judge the fluctuation of the real-time correction amount by using the filtered retention judgment value instead of counting as it is. This retention determination value gently follows the fluctuation of the real-time correction amount by causing a time difference by filtering processing such as temporary delay processing. If the retention determination value is on the rich side of the rich determination value set on the rich side, it is determined that the value of the second air-fuel ratio detecting means 14 is retained in the lean side. On the contrary, if the retention determination value is on the lean side of the lean determination value set on the lean side, it is determined that the value of the second air-fuel ratio detecting means 14 is richly retained. The integrated value of the lean residence time is a value obtained by counting the period during which the residence determination value is on the rich side of the rich determination value. The electronic control unit 30 continues to execute the residence time integration means 33, which adds the integrated value of the lean residence time in this way, while constantly monitoring.

The standard for changing the lean correction value is that the integrated value of the lean residence time exceeds a predetermined value. This is because the integrated value of the lean residence time is effective for determining the air-fuel ratio deviation failure between cylinders. It is desirable that the air-fuel ratios of the plurality of cylinders 2 are common to each other, but this may deviate significantly. This failure is called an air-fuel ratio deviation failure between cylinders (AFIM). When an air-fuel ratio deviation failure occurs between cylinders, the output of the first air-fuel ratio detecting means 12 detected immediately after the exhaust gas from each cylinder 2 merges tends to shift to the rich side. At this time, since the main feedback control performs lean correction, the catalyst of the exhaust gas purification unit 13 has a lean atmosphere. At this time, the value of the second air-fuel ratio detecting means 14 indicates a lean output. When this state continues, long-time learning is performed in which the long-time correction amount acting on the target air-fuel ratio is richly corrected in order to suppress the exhaust gas. This long-time learning is performed as a part of the sub-feedback control means 32 included in the electronic control unit 30.

The long time correction amount by this long time learning is increased as the integrated value of the lean residence time increases. However, if this long time correction amount becomes too large, the influence on the target air-fuel ratio becomes too large, and the real-time correction amount with the normal setting cannot be adjusted. In that case, the value of the second air-fuel ratio detecting means 14 becomes the rich side depending on the operating condition, and even if the correction must be made lean, the target air-fuel ratio by the main feedback control shifts to the rich side. Therefore, the lean correction value in the real-time correction amount is adjusted in a plurality of steps or steplessly based on the integrated value of the lean residence time. That is, the standby correction amount control means 35 that increases the lean correction value in the real-time correction amount as the integrated value of the lean residence time increases is executed. This change in the lean correction value is recorded in the electronic control unit 30 as a preparation value, and is not immediately reflected in the sub-feedback control. When the value of the second air-fuel ratio detecting means 14 is on the rich side of the rich predetermined value, that is, at the timing of the change in which the sub-feedback control means 32 actually makes a lean correction of the real-time correction amount, the value to be changed is It will be larger than the default value. As a result, even if the overcorrection of the long time correction amount by the long time learning is taken into consideration, the influence of the lean correction value by the real-time correction amount becomes large, and it can be brought closer to the lean side.

The standby correction amount control means 35 does not reflect the integrated value of the lean residence time in the long time correction amount when it is less than the predetermined variation determination value, and corrects it to the rich side when it exceeds the variation determination value. It is desirable to reflect it. Each engine 1 and each cylinder 2 has its own individuality, and if there is even a little time to stay in the lean, it may be inappropriate to reflect it in the long time correction amount, so until it starts to reflect it. It is desirable to have a surplus width.

On the other hand, when the situation actually improves, such as when the air-fuel ratio deviation failure between cylinders is improved, it is desirable that the integrated value of the lean residence time used for the determination can be subtracted. Therefore, the electronic control unit 30 executes the residence time subtraction means 36 for subtracting the integrated value of the lean residence time according to the residence time when the residence determination value stays on the lean side from the lean determination value.

When the integrated value of the lean residence time is subtracted and falls below a predetermined value, the long time correction amount is gradually reduced accordingly. This is because even if the situation improves, if the correction is continued to the extremely rich side, the exhaust gas cannot be suppressed.

However, if the amount of long-time correction is reduced even with a slight improvement, learning returns will occur frequently. Therefore, it is preferable to lower the lean determination value, which is a criterion for determining to subtract the integrated value of the lean residence time related to the decrease of the long time correction amount, as the integrated value of the lean residence time increases. The electronic control unit 30 executes the learning return deterrent means 37 that reduces the lean determination value stepwise or steplessly. Due to this decrease in the lean determination value, the time for determining that the residence time subtracting means 36 has retained on the lean side is reduced, and learning return is less likely to occur.

The standby correction amount control means 35 and the learning return suppression means 37 may be executed at the same timing when the integrated value of the lean residence time becomes a predetermined value or more. The predetermined value may be set in two stages, for example, a value for determining deterioration and a value for determining a higher failure, or may be set in multiple stages. Further, when the value becomes higher than the value for determining deterioration, the value may be increased steplessly.

An example of the flow of the electronic control unit 30 when an air-fuel ratio deviation failure between cylinders occurs will be described with reference to FIG. When an air-fuel ratio deviation failure between cylinders occurs (S101), the output of the first air-fuel ratio detecting means 12 is rich-shifted (S102). On the other hand, since the main feedback control means 31 works, the value of the second air-fuel ratio detecting means 14 stays lean. The electronic control unit 30 executes the residence time integration means 33, and the integrated value of the lean residence time is added (S103 → No → S104 → S111 → lean → S114). At this time, the sub-feedback control means 32 reflects the real-time correction amount in the rich correction value on the rich side (S115). In the initial stage, the integrated value of the lean residence time is less than the variation judgment value (return to S116 → No → S103), but when the lean residence time is prolonged and exceeds the variation judgment value (S116 → Yes), the long time correction amount is adjusted. It is corrected to the rich side (S117).

When the integrated value of the lean residence time becomes equal to or higher than the deterioration determination value higher than the variation determination value (S103 → Yes), the electronic control unit 30 executes the standby correction amount control means 35 and the learning return suppression means 37. (S105). The standby correction amount control means 35 increases the lean correction value of the real-time correction amount as compared with the normal state. Further, the learning return suppressing means 37 also increases the lean determination value of the retention determination value as compared with the normal state. However, these value changes are retained as reserve values inside the electronic control unit 30, and changes to the expanded values are not performed at this stage. While the value of the second air-fuel ratio detecting means 14 stays lean (S111 → lean), the situation continues (S114 to S117).

Further, when the integrated value of the lean residence time becomes equal to or higher than the failure determination value higher than the deterioration determination value (S103 → Yes), the standby correction amount control means 35 and the learning return suppression means 37 are further executed (S105). ), The lean correction value of the real-time correction amount is further expanded in multiple steps, and the lean determination value of the retention determination value is also expanded in multiple steps.

If the value of the second air-fuel ratio detecting means 14 is neither lean nor rich, the real-time correction amount is ± 0 (S111 → intermediate → S112).

If the value of the second air-fuel ratio detecting means 14 becomes rich after the standby correction amount control means 35 and the learning return suppressing means 37 in S105 are executed in one step or more (S105 → S111 → rich), the step becomes one or more steps. The real-time correction amount with the increased rich correction value is reflected on the lean side (S113). As a result, the long time correction amount can be offset to the rich side or more to the lean side.

Although not shown, if the value of the second air-fuel ratio detecting means 14 continues to be rich, the staying time subtracting means is used even if the staying determination value is expanded to the lean side by the learning return suppressing means 37 in S105. 36 will be executed after S113.

An example of transition of the set value when the control device according to the present invention is executed will be described with reference to FIG. The situation is that an air-fuel ratio deviation failure between cylinders has occurred, and the value of the first air-fuel ratio detecting means 12 is rich-shifted, so that the main feedback control means is controlled so as to be directed toward lean. Therefore, the value of the second air-fuel ratio detecting means 14 is basically closer to the lean side. If the value of the second air-fuel ratio detecting means 14 is on the lean side of the lean predetermined value set on the lean side, the real-time correction amount related to the target air-fuel ratio is set to the rich correction value (T01). The lean predetermined value and the rich predetermined value set on the rich side are both set as determination voltages to be compared with the voltage value of the sensor in the processing of the electronic control unit 30, but are described here as predetermined values. During this period (T01 to T05), the integrated value of the lean residence time is added as described later, and the long time correction amount becomes rich. When the value of the second air-fuel ratio detecting means 14 becomes richer than the lean predetermined value, the value of the real-time correction amount is returned to ± 0 from the rich correction value (T04). Further, when the value of the second air-fuel ratio detecting means 14 temporarily becomes richer than the rich predetermined value due to operating conditions or other reasons, the value of the real-time correction amount is set to the lean correction value (T06 to T08). ). The lean correction value set at this time is the value at the normal time. After that, the value of the second air-fuel ratio detecting means 14 continues to be on the lean side of the lean predetermined value (T10 to T14). At this time as well, the integrated value of the lean residence time is added as described later, and the long time correction amount is also rich. After that, due to other circumstances such as a change in the operating condition, the condition is improved and the value of the second air-fuel ratio detecting means 14 becomes rich (T16 ~). We will explain how each value fluctuates when the above transitions occur.

The real-time correction amount is immediately reflected on the rich side when the value of the second air-fuel ratio detecting means 14 is closer to the lean side than the lean predetermined value (T01). However, the retention judgment value obtained by filtering the real-time correction amount gradually moves toward the rich side so as to follow this. Therefore, the retention determination value is on the rich side of the rich determination value for which the rich determination is made from T02, which has a slight time lag from T01. Further, the retention determination value is not richly determined (T05) after the real-time correction amount returns to 0 after T04. Further, when the value of the second air-fuel ratio detecting means 14 becomes rich and the real-time correction amount is changed to the lean correction value (T06), the retention determination value becomes leaner than the lean determination value after a time lag (T07).

From T02 where the residence determination value is on the rich side of the rich determination value, the addition of the integrated value of the lean residence time is started. However, until this addition reaches the variation determination value (T02 to T03), the long time correction amount remains 0 and is not changed. When the integrated value of the lean residence time reaches the variation determination value, the sub-feedback control means enriches the long time correction amount (T03). As a result, the target air-fuel ratio is doubled toward the rich side by adding the rich correction value of the real-time correction amount and the value at the time of determining the variation of the long time correction amount (T03 to T04). On the other hand, at T07 to T09 where the residence determination value is on the lean side of the lean determination value, the integrated value of the lean residence time is subtracted. After that, in the intermediate state where the residence determination value is neither lean nor rich, the integrated value of the lean residence time does not increase or decrease (T09 to T11). The residence determination value becomes richer than the rich determination value (T11), and the integrated value of the lean residence time is added (T11 to 1) after the real-time correction amount becomes T10, which is the rich correction value. The increase / decrease in the target air-fuel ratio during this period (T03 to T12) reflects the increase / decrease in the real-time correction amount.

When the lean residence time exceeds the deterioration judgment value, the long time correction amount is enriched (T12). Further, the first step of the standby correction amount control means 35 and the learning return suppression means 37 is executed here. The standby correction amount control means 35 changes the lean correction value indicated by the thick broken line of the real-time correction amount from the value at the normal time to the value at the time of deterioration. This change is preliminary, and at this stage, the value of the real-time correction amount itself is not changed, but the value that will be reserved for the next change. Similarly, the learning return suppressing means 37 changes the lean determination value indicated by the thick broken line of the retention determination value from the value at the time of normal to the value at the time of deterioration. This change is also preliminary, and at this stage, the value of the retention determination value itself is not changed, but the value that will be reserved when the next change is made.

Further, when the lean residence time exceeds the failure determination value set higher than the deterioration determination value, the long time correction amount is further enriched (T13). Here, the second stage of the standby correction amount control means 35 and the learning return suppression means 37 is further executed. The standby correction amount control means 35 changes the lean correction value from the value at the time of deterioration to the value at the time of failure, which is lean. The learning return suppressing means 37 changes the lean determination value from the value at the time of deterioration to the value at the time of failure, which is lean.

After that, when the value of the second air-fuel ratio detecting means 14 becomes an intermediate value on the rich side of the lean predetermined value (T14), the real-time correction amount is changed from the rich correction value to 0. After a time lag from that point, the residence determination value becomes less than the rich determination value (T15), and the addition of the integrated value of the lean residence time stops. However, in this example, since the integrated value of the lean residence time has reached the upper limit at the stage of T14 before that, the upper limit remains.

Then, when the value of the second air-fuel ratio detecting means 14 becomes richer than the rich determination value (T16), the real-time correction amount is changed to the lean correction value. At this time, since the lean correction value is set to the lean side by the standby correction amount control means first, the correction range of the real-time correction amount to the lean side becomes large. As a result, even if the long time correction amount stays high on the rich side, the real-time correction amount becomes sufficiently lean, so the target air-fuel ratio is not the value of the broken line in the figure, but the thick solid line that is leaner than that. It behaves like this. As a result, it becomes possible to sufficiently cope with the situation where the value of the second air-fuel ratio detecting means 14 becomes rich.

On the other hand, the retention judgment value moves from the real-time correction amount to the lean side through a time lag, but since the lean judgment value is a value at the time of failure that is leaner than the normal value, the retention judgment value is lean. A sufficient time (T16 to T17) is required to reach the lean determination value. Only after waiting until T17, it is determined that the residence determination value is lean, and the residence time subtraction means 36 for subtracting the integrated value of the lean residence time is executed. When the subtraction progresses and the integrated value of the lean residence time falls below the value of the failure determination, the value of the long time correction amount becomes slightly lean from the rich side (T18). At this time, the long-time learning remains rich in the long-time learning, but the target air-fuel ratio is corrected to the lean side because the lean correction value of the real-time correction amount is the value at the time of deterioration.

When the above deterioration determination and failure determination are made, the electronic control unit 30 may appropriately modify the fuel injection conditions.

1 Engine 2 Cylinder 3 Throttle valve 4 Intake passage 5 Exhaust passage 10 In-cylinder injection device 12 First air-fuel ratio detection means 13 Exhaust purification unit 14 Second air-fuel ratio detection means 15 Muffler 30 Electronic control unit 31 Main feedback control means 32 Sub-feedback Control means 33 Resident time integration means 35 Standby correction amount control means 36 Resident time subtraction means 37 Learning return suppression means

Claims (6)

An injection device that supplies fuel to an engine with multiple cylinders,
A first air-fuel ratio detecting means provided in the exhaust passages drawn from the plurality of cylinders and detecting the air-fuel ratio in the exhaust passages,
An exhaust gas purification device provided on the downstream side of the first air-fuel ratio detecting means,
A second air-fuel ratio detecting means provided in the exhaust passage on the downstream side of the exhaust purification device and detecting the air-fuel ratio in the exhaust passage,
Based on the information obtained by the first air-fuel ratio detecting means, a main feedback control means that feedback-controls the fuel injection amount by the injection device so that the air-fuel ratio becomes a preset target air-fuel ratio.
If the information obtained by the second air-fuel ratio detecting means is leaner than the lean predetermined value set on the lean side, the target air-fuel ratio may be rich-corrected and richer than the rich predetermined value set on the rich side. For example, a long for correcting the real-time correction amount for lean correction of the target air-fuel ratio and the integrated value of the lean residence time obtained based on the information obtained by the second air-fuel ratio detecting means. Sub-feedback control means for setting the time correction amount and
With
When an air-fuel ratio deviation failure occurs between cylinders in which an air-fuel ratio deviation occurs between the plurality of cylinders, the sub-feedback control means sets a lean correction value, which is a value for lean correction of the real-time correction amount, to stay in the lean state. An engine control device that adjusts in multiple steps or steplessly based on the integrated value of time. The engine control device according to claim 1, wherein the integrated value of the lean residence time is calculated based on the total amount of time that the residence determination value obtained by filtering the real-time correction amount stays at a predetermined value on the rich side. A standby correction amount control means that holds the lean correction value as a reserve value and applies the reserve value at the time of subsequent acceleration.
The engine control device according to claim 1 or 2. The long time correction amount is not reflected when the integrated value of the lean residence time is less than a predetermined variation determination value, and reflects the correction to the rich side when the variation determination value is exceeded.
The engine control device according to any one of claims 1 to 3. The residence time subtracting means for subtracting the integrated value of the lean residence time according to the residence time at which the residence determination value stays on the lean side from the lean determination value is executed.
The engine control device according to any one of claims 1 to 4. The learning return deterrent means for lowering the lean determination value according to the integrated value of the lean residence time is executed.
The engine control device according to claim 5.

Images